How a network recognises an image
In short: Filters find edges, deeper layers combine them into shapes, and the network scores each possible label.
1. The image goes in as a grid of pixel values. 2. Small filters slide across the image, each responding to a pattern such as an edge. 3. Deeper layers combine the simple patterns into more complex shapes. 4. The final layers give a score for each possible label. 5. The label with the highest score is the network's answer.
Sources: Wikipedia, “Convolutional neural network (Wikipedia)” (CC BY-SA 4.0). A simplified picture of a convolutional network.
Opens the Intellect+ feed with this card first. Swipe up for the next one.
Sources
- Summary by Intellect · source: Convolutional neural network
To cite this card, link to https://intellectplus.co.uk/app/how-a-neural-network-recognises-images.
How we source, check and correct cards: Editorial standards.
General information, not professional advice. Spotted an error? Flag the card or see corrections and complaints.
Intellect Plus is a visual educational discovery platform for exploring fascinating facts and ideas. Brain Food, Not Brain Rot. What is Intellect Plus?
